CN113379211A - Block chain-based logistics information platform default risk management and control system and method - Google Patents
Block chain-based logistics information platform default risk management and control system and method Download PDFInfo
- Publication number
- CN113379211A CN113379211A CN202110605907.0A CN202110605907A CN113379211A CN 113379211 A CN113379211 A CN 113379211A CN 202110605907 A CN202110605907 A CN 202110605907A CN 113379211 A CN113379211 A CN 113379211A
- Authority
- CN
- China
- Prior art keywords
- logistics
- data
- default risk
- default
- block chain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012502 risk assessment Methods 0.000 claims abstract description 52
- 238000012954 risk control Methods 0.000 claims abstract description 20
- 238000011156 evaluation Methods 0.000 claims abstract description 15
- 238000013210 evaluation model Methods 0.000 claims abstract description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 16
- 238000003066 decision tree Methods 0.000 claims description 14
- 238000003860 storage Methods 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 13
- 238000005070 sampling Methods 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000007637 random forest analysis Methods 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 6
- 238000005295 random walk Methods 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 6
- 238000007726 management method Methods 0.000 description 25
- 230000008569 process Effects 0.000 description 13
- 238000012384 transportation and delivery Methods 0.000 description 6
- 238000011176 pooling Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012854 evaluation process Methods 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 238000013439 planning Methods 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 101100119135 Mus musculus Esrrb gene Proteins 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Game Theory and Decision Science (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a block chain-based logistics information platform default risk management and control system and a block chain-based logistics information platform default risk management and control method, wherein the system comprises the following components: the block chain platform is used for storing logistics data uploaded by each logistics network; a breach risk assessment system for: acquiring logistics data of each logistics network from a block chain platform, constructing a default risk evaluation model, and evaluating default risks of each logistics network through the default evaluation model to obtain default risk scores of each logistics network; and the default risk management and control system is used for acquiring default risk scores of the logistics nodes generated by the default risk evaluation system and carrying out classified default risk management and control on the logistics nodes based on the default risk scores. According to the method, the default risk assessment of each logistics network is finished by directly adopting the logistics data uploaded by each logistics network stored on the block chain platform, and the classified risk control of each logistics network is implemented based on the assessment result, so that the authenticity and fairness of the assessment result are ensured, and the risk control effect of each logistics network is improved.
Description
Technical Field
The invention relates to the field of logistics, in particular to a block chain-based logistics information platform default risk management and control system and method.
Background
The logistics transportation is a composite service industry integrating multiple links of transportation, storage, distribution, sale and the like. In recent years, with the continuous development of the internet and information technology, the traditional logistics mode can not meet the market demand, logistics information technology is generally introduced into logistics enterprises, information generated in the logistics process is collected, stored, collected and analyzed, and the intelligent and automatic levels of a logistics system are continuously improved.
The logistics process involves numerous links, and the problems of production data counterfeiting, information leakage, quality safety and the like are difficult to avoid. Therefore, some large logistics enterprises try to introduce a block chain platform into a logistics information platform, and key logistics information of each logistics network is linked and stored to the block chain platform, so that non-tampering storage of logistics data is achieved.
Due to the fact that logistics points involved in the logistics process are multiple, the process is complex, and the types of default risks are various, the default risk assessment of each logistics point is achieved, and the risk management and control of each logistics point according to default risk assessment results become problems to be solved urgently in the industry. At the present stage, the default risk of each logistics network is generally evaluated by a professional evaluation organization, and both the evaluation process and the evaluation result can be subjected to human intervention and tampering, so that the authenticity and the effectiveness of the evaluation result are reduced, and finally the effect of risk control of each logistics network is poor.
The invention aims to provide a default risk assessment and control strategy, which directly adopts logistics data uploaded by each logistics network point stored on a block chain platform to complete default risk assessment of each logistics network point, and implements risk control of each logistics network point based on an assessment result, thereby ensuring authenticity and fairness of the assessment result and finally improving risk control effect of each logistics network point.
Disclosure of Invention
In order to achieve the above technical objective of the present invention, a first aspect of the present invention provides a block chain-based logistics information platform default risk management and control system, which has the following specific technical solutions:
a logistics information platform default risk management and control system based on block chains comprises:
the block chain platform is used for storing logistics data uploaded by each logistics network;
a breach risk assessment system for:
acquiring logistics data of each logistics network from the block chain platform; and
constructing a default risk assessment model based on the logistics data of each logistics network, and assessing the default risk of each logistics network through the default assessment model to obtain a default risk score of each logistics network;
and the default risk management and control system is used for acquiring the default risk scores of the logistics outlets generated by the default risk assessment system and performing classified default risk management and control on the logistics outlets based on the default risk scores of the logistics outlets.
In some embodiments, the blockchain platform comprises a data acquisition module, a data processing module, and a blockchain storage system, wherein: the data acquisition module is used for acquiring the logistics data uploaded by each logistics network; the data processing module is used for realizing pre-uplink preprocessing of the collected logistics data, and the pre-uplink preprocessing comprises one or more of data cleaning, missing value processing, abnormal value processing and duplicate removal processing; and the block chain storage system is used for realizing permanent and tamper-free storage of the logistics data after the preprocessing is finished.
In some embodiments, the breach risk assessment system comprises a data acquisition module, a model building module, and an assessment module, wherein: the data acquisition module is used for acquiring logistics data of each logistics network from the block chain platform; the model construction module is used for constructing a default risk assessment model based on the acquired logistics data of each logistics network; and the evaluation module is used for evaluating the default risk of each logistics network point through the default evaluation model so as to obtain the default risk score of each logistics network point.
In some embodiments, the logistics data includes historical order data, historical credit characteristics, people credit characteristics, and real-time business data, wherein the historical order data includes formatted order data and unformatted order data. The step of constructing a default risk assessment model based on the logistics data of each logistics network comprises the following steps: extracting first default risk data characteristics related to default from the formatted order data by adopting a random forest algorithm; extracting a second type of default risk data characteristics related to default from the unformatted order data by adopting a deep learning algorithm; and constructing a default risk assessment model based on the first default risk data characteristic, the second default risk data characteristic, the historical credit characteristic, the character credit characteristic and the real-time business data of each logistics network point.
In some embodiments, said extracting a first type of breach risk data characteristic related to breach from said formatted order data using a random forest algorithm comprises: acquiring a sample data set comprising a plurality of formatted order data samples; randomly extracting m training samples from the sample data set in a place where the sample data set is placed back by using a bootstrapping sampling method, and performing n rounds of extraction to obtain n training sets; respectively training n decision tree models based on n training sets; respectively calculating the importance value of each feature of the formatted order data by using n decision tree models, and averaging the n importance values of each feature to obtain a unique determined importance value of each feature; selecting a plurality of characteristics of which the importance value is determined to exceed a predetermined threshold value as the first type of default risk data characteristics.
In some embodiments, said extracting a second type of breach risk data characteristic related to the breach from the unformatted order data using a deep learning algorithm comprises: converting the unformatted order data into a plurality of word vectors; constructing a feature matrix based on the word vectors; and performing feature extraction on the feature matrix by adopting a convolutional neural network to obtain the second-class default risk data features.
In some embodiments, said constructing a default risk assessment model based on said first type of default risk data feature, said second type of default risk data feature, said historical credit feature, said persona credit feature, and said real-time business data comprises: a component graph neural network G ═ V, E, where V is a set of nodes and E is a set of edges; adding each logistics network point into a node set as a node, wherein the first type default risk data characteristic, the second type default risk data characteristic, the historical credit characteristic, the character credit characteristic and the real-time service data of each logistics network point are used as the data characteristics of the corresponding node; and acquiring a logistics transportation record from the block chain platform to acquire the relation among the logistics network points, adding the acquired relation among the logistics network points into the edge set, and acquiring the default risk assessment model based on the graph neural network.
In some embodiments, said evaluating the default risk of each of the logistics outlets by the default evaluation model to obtain a default risk score for each of the logistics outlets comprises: selecting nodes with default risk labels as seed nodes, and executing a random walk algorithm in the default risk evaluation model by adopting a Node2VEC algorithm to obtain the sampling probability of each Node; and obtaining the corresponding risk score of each logistics network point based on the sampling probability of each node.
In some embodiments, the classifying the breach risk management of each logistics site based on the breach risk score of each logistics site comprises: for the first type of logistics network points with default risk scores lower than a preset threshold value, adopting a first type of risk control measures; and taking a second type of risk control measures for the second type of logistics network points with default risk scores equal to or higher than a preset threshold value.
The second aspect of the present invention provides a block chain-based method for managing and controlling a breach risk of a logistics information platform, which includes:
acquiring logistics data uploaded by each logistics network from the block chain platform, wherein the logistics data comprise historical order data, historical credit characteristics, character credit characteristics and real-time service data;
constructing a default risk assessment model based on the logistics data of each logistics network point acquired from the block chain platform, and assessing the default risk of each logistics network point through the default assessment model to obtain a default risk score of each logistics network point;
and acquiring default risk scores of the logistics outlets generated by the default risk assessment system, and performing classified default risk control on the logistics outlets based on the default risk scores of the logistics outlets.
Therefore, the system and the method for managing and controlling the default risk of the logistics information platform directly adopt the logistics data uploaded by each logistics network point stored on the blockchain platform to complete the default risk assessment of each logistics network point, and implement the classified risk management and control of each logistics network point based on the assessment result, so that the authenticity and the fairness of the assessment result are ensured, and the risk management and control effect of each logistics network point is improved.
In addition, the logistics data used in the default risk assessment comprise historical order data of logistics nodes and data of multiple dimensions such as historical credit characteristics, character credit characteristics and real-time business data of all the logistics nodes, the richness of the data is greatly improved, and finally the accuracy of the default risk assessment result is greatly improved.
Drawings
Fig. 1 is a block diagram of a block chain-based logistics information platform default risk management and control system according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating the default risk assessment system in the embodiment of the present invention constructing a default risk assessment model.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Due to the fact that logistics points involved in the logistics process are multiple, the process is complex, and the types of default risks are various, the default risk assessment of each logistics point is achieved, and the risk management and control of each logistics point according to default risk assessment results become problems to be solved urgently in the industry.
At the present stage, the default risk of each logistics network is generally evaluated by a professional evaluation organization, and both the evaluation process and the evaluation result can be subjected to human intervention and tampering, so that the authenticity and the effectiveness of the evaluation result are reduced, and finally the effect of risk control of each logistics network is poor.
The block chain-based logistics information platform default risk management and control system and method provided by the aspect aim to solve the technical problems in the prior art.
The technical solution of the present invention and how to solve the above technical problems will be described in detail with specific embodiments below. The following embodiments may be combined, and the same or similar concepts or processes may not be described in detail in some embodiments.
Embodiments of the present invention will be described below with reference to the accompanying drawings.
Example one
As shown in fig. 1, the logistics information platform default risk management and control system based on a block chain provided in the embodiment of the present invention includes a block chain platform 10, a default risk assessment system 20, and a default risk management and control system 30, where:
and the block chain platform 10 is used for storing the logistics data uploaded by each logistics network.
A breach risk assessment system 20 for:
acquiring logistics data of each logistics network from the block chain platform 10; and
and constructing a default risk evaluation model based on the logistics data of each logistics network, and evaluating the default risk of each logistics network through the default evaluation model to obtain the default risk score of each logistics network.
And the default risk management and control system is used for acquiring default risk scores of the logistics nodes generated by the default risk evaluation system and carrying out classified default risk management and control on the logistics nodes based on the default risk scores of the logistics nodes.
The default risk management and control system 30 is configured to obtain default risk scores of the logistics nodes generated by the default risk assessment system 20, and perform classified default risk management and control on the logistics nodes based on the default risk scores of the logistics nodes.
With continued reference to fig. 1, optionally, the blockchain platform 10 includes a data acquisition module 11, a data processing module 12, and a blockchain storage system 13, wherein:
and the data acquisition module 11 is used for acquiring the logistics data uploaded by each logistics network.
And the data processing module 12 is configured to implement pre-processing on the acquired logistics data before uplink. Optionally, the preprocessing includes data cleaning, missing value processing, abnormal value processing, deduplication processing, and the like.
And the block chain storage system 13 is used for realizing permanent and tamper-free storage of the logistics data after the preprocessing is finished.
With continued reference to fig. 1, optionally, breach risk assessment system 20 includes a data acquisition module 21, a model building module 22, and an assessment module 23, wherein:
and the data acquisition module 21 is configured to acquire logistics data of each logistics node from the blockchain platform 10.
And the model building module 22 is configured to build a default risk assessment model based on the acquired logistics data of each logistics network.
And the evaluation module 23 is configured to evaluate the default risk of each logistics site through the default evaluation model to obtain a default risk score of each logistics site.
The logistics information platform default risk control system provided by this embodiment directly adopts the logistics data uploaded by each logistics site stored on the blockchain platform to complete the default risk assessment of each logistics site, and implements the classification risk control of each logistics site based on the assessment result, thereby ensuring the authenticity and fairness of the assessment result, and finally improving the risk control effect of each logistics site.
Example two
In this embodiment, the logistics data uploaded by each logistics node includes historical order data, historical credit characteristics, character credit characteristics, and real-time service data, where the historical order data includes formatted order data and unformatted order data according to a specific storage format of the data. Wherein:
formatted order data refers to row data comprising a plurality of characteristic values (attributes), which can be saved into a two-dimensional table or relational database. Unformatted order data refers to data held in a textual format.
In this embodiment, the formatted order data includes the following features: the basic data characteristics specifically comprise order numbers, delivery sites, delivery member information, collecting sites, express information, delivery information and the like. The service data characteristics specifically comprise the business qualification of the network points, income conditions, available storage area, network point owing, single ticket penalty, daily average traffic, port delivery timeliness, departure checkpoint rate, current day label yield, false acceptance complaint rate, loss damage rate, upgrade complaint acceptance complaint rate, timely acceptance rate, total complaint rate, secondary complaint rate, one-time solution rate of the whole channel, acceptance rate and the like. And (4) historical default records of the network points, wherein the attributes can be used as default risk labels of corresponding formatted order data. As is familiar to those of ordinary skill in the art, there is an association between business data characteristics and the risk of breach at a logistics site.
In this embodiment, the unformatted order data includes text data such as item description information, customer evaluation information, customer complaint information, and the like.
In this embodiment, the historical credit features include one or more of financial status, records complained, records of default, records related to litigation, and records of administrative penalties. The character credit characteristics comprise one or more of operation condition, asset condition, storage area, transport vehicle, personnel number and informatization level. The real-time service data comprises one or more of line achievement rate, timely item collecting rate, service good evaluation rate, abnormal problem solving rate, information timeliness rate, information integrity rate and information accuracy rate.
As shown in fig. 2, in the present embodiment, the default risk assessment system 20 constructs a default risk assessment model as follows.
Firstly, data acquisition:
that is, the data obtaining module 21 obtains logistics data of logistics sites of each logistics site stored on the blockchain platform, where the logistics data includes historical order data, historical credit characteristics, character credit characteristics, and real-time service data, and the historical order data includes formatted order data and unformatted order data.
Second, extracting the first type of default risk data features:
that is, the model building module 22 extracts the first type of default risk data features related to the default from the formatted order data by using a random forest algorithm.
As indicated above, the formatted order data includes a large number of business data features, and as those skilled in the art will appreciate, not every feature has a large correlation with the risk of default, and therefore, it is necessary to select an appropriate number of features from these features that have a high correlation with the risk of default. The selected characteristics are the first type default risk data characteristics.
Therefore, optionally, the specific process of extracting the first type of default risk data features related to the default from the formatted order data by the model building module 22 using the random forest algorithm is as follows:
step 1, obtaining a sample data set comprising a plurality of formatted order data samples.
And 2, randomly extracting m training samples from the sample data set by using a Bootstrap sampling method, and performing n rounds of extraction to obtain n training sets.
For example, in some alternative embodiments, the sample size of the sample data set is 900, for a total of 10 rounds (n-10) with replacement decimation. Each round of extraction is specifically as follows:
600(m ═ 600) training samples are extracted from the sample data set to serve as a training set formed by the extraction of the current round, and the other 300 sample data which are not extracted are recorded as an out-of-bag data set formed by the extraction of the current round.
After 10 rounds of extraction are completed, 10 training sets are obtained in total, and 10 out-of-bag data sets are correspondingly obtained.
And 3, respectively training n decision tree models based on n training sets.
In this embodiment, optionally, 10 training sets are respectively used to complete the training of the decision tree model, so that 10 trained decision tree models can be obtained, and the 10 decision tree models also form the random forest model in this embodiment.
And 4, respectively calculating the importance value of each feature of the formatted order data by using the n decision tree models, and averaging the n importance values of each feature to obtain a unique determined importance value of each feature.
The importance values of the features of the formatted order data are calculated using n (10 as in the above-described embodiment) decision tree models, respectively. Specifically, the method comprises the following steps:
each decision tree model selects the corresponding out-of-bag data set to calculate out-of-bag data errors, namely the prediction error rate of the corresponding decision tree model is calculated by using the out-of-bag data set, the result is recorded as err1, then, noise influence is added to the characteristic x of the sample data in the out-of-bag data set, for example, the value of the characteristic x is changed randomly, and the out-of-bag data errors are calculated again and recorded as err 2. Then the importance value of feature x calculated by the decision tree model can be characterized as err2-err 1.
It can be seen that for each feature x, n decision tree models calculate n importance values, and thus the final determined importance value for feature x is determined by averaging. The determined importance value is:
∑(err2-err1)/n；
after the above processing, the determination importance values of all the features are calculated.
And 5, selecting a plurality of characteristics with the determined importance value exceeding a preset threshold value as first-class default risk data characteristics.
Optionally, all the features are sorted in a descending order according to the determined importance value, and a plurality of the features arranged in the front are selected as the first type of default risk data features.
For example, in the present embodiment, the top 17 features are selected as the first type of default risk data features. Optionally, the 17 features are classified into three categories according to their specific meanings, specifically:
the network management condition is as follows: including the business qualifications, income conditions, available warehousing areas, network owers, and average daily traffic of the franchised network sites.
The network node operation condition is as follows: including the time rate of departure delivery, the rate of departure punctuality, the yield of the tag on the day, the loss breakage rate and the timely yield.
The service condition of the network points is as follows: false complaint rate of sign-off, upgraded complaint acceptance complaint rate, total complaint rate, secondary complaint rate, one-time solution rate of all channels, rating rate of sign-off and rating rate of collection.
At this point, the formatted order data is subjected to machine learning, and business data features with high correlation degree with default risks, namely, default risk data features of the first type, are extracted.
Thirdly, extracting the characteristics of the second type of default risk data:
that is, the model building module 22 adopts a deep learning algorithm to extract the second type of default risk data features related to the default from the unformatted order data.
As described above, unformatted order data is text data, and a deep learning algorithm is required to mine default risk data characteristics hidden in the text data and related to the default.
Optionally, in this embodiment, the model building module 22 completes extraction of the second type of default risk data features by using a Convolutional Neural Network (CNN), and the specific steps are as follows:
step 1, converting the unformatted order data into a plurality of word vectors.
Optionally, Word vector conversion may be performed on the unformatted order data by using Word vector algorithms such as Word2Vec and Glo Ve to obtain a plurality of Word vectors, where each Word vector corresponds to a keyword in the unformatted order data. That is, the unformatted order data is characterized as a number of word vectors.
And 2, constructing a feature matrix based on the word vectors.
That is, all the word vectors are stored in a matrix, and each word vector corresponds to one column of the feature matrix.
And 3, extracting the features of the feature matrix by adopting a convolutional neural network to obtain a second type of default risk data features.
It is a conventional technique in the art to extract hidden features by using a convolutional neural network, which generally comprises a convolutional layer, a pooling layer, a ReLU layer, and a fully-connected layer, wherein:
and (3) rolling layers: the method comprises two key operations, namely local correlation and window sliding, multiplying different local matrixes of a convolution layer word vector matrix by each position element of a convolution kernel matrix, and then adding to complete convolution operation to obtain a new characteristic matrix.
A pooling layer: and compressing each submatrix of the input tensor to obtain a value, and reducing the dimension of the input matrix, wherein the average value of the corresponding area is used as the element value after pooling.
Relu layer: the correction linear unit corrects the output to be a negative value and 0, and does not change when the output is a positive value.
Full connection layer: and splicing the feature matrixes calculated by the convolutional layer and the pooling layer into a one-dimensional matrix.
After the feature extraction, the finally obtained second-class default risk data features include: and the security of express items, the false signature yield, the human good comment behavior, the false operation condition data and other default characteristics.
At this point, by performing deep mining on the unformatted order data, default features related to the default, namely, the second type of default risk data features, are extracted.
It should be noted that the execution sequence of the first type of default risk data feature extraction and the second type of default risk data feature extraction may be exchanged, and after the second type of default risk data feature extraction is executed first, the first type of default risk data feature extraction is executed again. Of course, the first type of default risk data feature extraction and the second type of default risk data feature extraction may also be performed in parallel.
Fourthly, constructing a default risk assessment model:
that is, the model building module 22 builds the default risk assessment model based on the first type default risk data characteristics, the second type default risk data characteristics, the historical credit characteristics, the character credit characteristics and the real-time business data of each logistics network.
Optionally, the specific steps of the model building module 22 building the default risk assessment model are as follows:
step 1, a component diagram neural network G is (V, E), wherein V is a node set, and E is an edge set.
And 2, adding each logistics network point into the node set as a node, wherein the first type default risk data characteristics, the second type default risk data characteristics, the historical credit characteristics, the character credit characteristics and the real-time service data of each logistics network point are used as the data characteristics of the corresponding node.
And 3, acquiring logistics transportation records from the block chain platform to acquire the relation among the logistics network points, and adding the acquired relation among the logistics network points into an edge set to acquire a default risk assessment model based on the graph neural network.
Specifically, the method comprises the following steps: and acquiring data of an order number, a collecting network point, an originating center, a transfer center, a destination center, a delivery network point and the like in the logistics transportation process from the block chain platform, and determining whether the logistics network points in the logistics transportation process of the order are linked or not according to the order number.
And after the construction of the default risk assessment model is completed, training the model by random walk. Specifically, a Node with a default risk label (i.e., a Node having a history default record) is used as a seed Node, and the existing Node2VEC model is used for random walk to obtain a default risk score of each logistics Node. The specific walk strategy is as follows: for the current node t, the next selected sampling node x, the transition probability is calculated by the following formula:
wherein: p is the probability of return, and if p > max (q,1), then the sample will try not to go back, i.e. the next node is unlikely to be the previous node t. q is an access parameter, and if q >1, then the wandering will tend to run between nodes around the start point, which may reflect the breadth-first search (BFS) characteristics of a node. If q <1, the wander will tend to run farther, reflecting the Depth First Search (DFS) feature.
In one embodiment, p is 1, q is 2, the random walk path is continuously simulated according to the random walk rule until the sampling probability of each node in the whole graph neural network is stabilized, and the simulation is stopped, so that the sampling probability of each node can be characterized as the default risk of the corresponding logistics network point.
When the model is formally used for default risk assessment, sample data to be queried is input into the default risk assessment model and the trained graph neural network, and the stable sampling probability values on the corresponding nodes generated in the graph neural network are the default risks of the sample to be queried.
Since the default risk directly output by the neural network of the graph is a decimal less than 1, the default risk of the logistics network is more conveniently understood by a user. Optionally, the sampling probability value of each node is multiplied by 100 to serve as default risk score of the logistics network point corresponding to the node, so that the default risk score of each network point is in a range from 0 to 100, and the higher the score is, the greater the default risk is.
As can be seen from the above description, the logistics data used in this embodiment includes historical order data of the logistics sites, and data of multiple dimensions, such as historical credit features, character credit features, and real-time business data, of each logistics site to construct the default risk assessment model. The richness of the data is greatly improved, and finally, the accuracy of default risk evaluation results obtained by the default risk evaluation model is greatly improved.
EXAMPLE III
In this embodiment, after obtaining the default risk scores of the logistics sites generated by the default risk assessment system 20, the default risk management and control system 30 implements classification default risk management and control on the logistics sites in the following manner:
and for the first type of logistics network points with default risk scores lower than a preset threshold value, determining that the default risk is lower, and adopting first type of risk management and control measures. The risk control standard of the first type of risk control measure is low, for example, when a logistics plan (including logistics path planning, logistics network point recommendation, and the like) is implemented, it may be considered to increase the probability that the first type of logistics network point is selected.
And for the second type of logistics network points with default risk scores equal to or higher than the preset threshold, the default risk is determined to be higher, and second type of risk control measures are taken. The risk control standard of the second type of control is higher, for example, when a logistics plan (such as logistics path planning and logistics network point recommendation) is implemented, the probability that the second type of logistics network point is selected can be considered to be reduced. For another example, in the implementation process of logistics transportation, real-time key monitoring on the logistics process of the second type of logistics network points can be selected, and when default abnormality occurs in the second type of logistics network points, alarm information is timely produced.
In other embodiments, in order to make classification default risk management and control more effective and accurate, each logistics site can be classified into three or four categories according to default risk scores.
Further, in this embodiment, after obtaining the default risk scores of the logistics nodes generated by the default risk evaluation system 20, the default risk management and control system 30 realizes recommendation of the logistics nodes with low risk scores to a greater extent through collaborative filtering recommendation based on the default risk scores of the logistics nodes.
The collaborative filtering recommendation generates recommendation results by binding a user preference model with a risk score-based collaborative filtering algorithm. The execution process is as follows:
firstly, historical logistics ordering information of logistics users in a database and risk scores of all logistics network points are read to conduct preference analysis of the users to establish a customer model.
Then, the user preference of unknown services is predicted based on the designed logistics distribution service recommendation algorithm, and a recommendation result set which is possibly interested by the user is generated.
And finally, screening the recommendation result set generated by the algorithm based on a certain criterion to enable the recommendation result to be more accurate, and then forming a final logistics network recommendation list.
The invention has been described above with a certain degree of particularity. It will be understood by those of ordinary skill in the art that the description of the embodiments is merely exemplary and that all changes that come within the true spirit and scope of the invention are desired to be protected. The scope of the invention is defined by the appended claims rather than by the foregoing description of the embodiments.
Claims (10)
1. The utility model provides a commodity circulation information platform risk of default management and control system based on block chain which characterized in that, it includes:
the block chain platform is used for storing logistics data uploaded by each logistics network;
a breach risk assessment system for:
acquiring logistics data of each logistics network from the block chain platform; and
constructing a default risk assessment model based on the logistics data of each logistics network, and assessing the default risk of each logistics network through the default assessment model to obtain a default risk score of each logistics network;
and the default risk management and control system is used for acquiring the default risk scores of the logistics outlets generated by the default risk assessment system and performing classified default risk management and control on the logistics outlets based on the default risk scores of the logistics outlets.
2. The block chain-based logistics information platform default risk management and control system of claim 1, wherein the block chain platform comprises a data acquisition module, a data processing module and a block chain storage system, wherein:
the data acquisition module is used for acquiring the logistics data uploaded by each logistics network;
the data processing module is used for realizing pre-uplink preprocessing of the collected logistics data, and the pre-uplink preprocessing comprises one or more of data cleaning, missing value processing, abnormal value processing and duplicate removal processing;
and the block chain storage system is used for realizing permanent and tamper-free storage of the logistics data after the preprocessing is finished.
3. The block chain-based logistics information platform default risk management and control system of claim 1, wherein the default risk assessment system comprises a data acquisition module, a model construction module and an assessment module, wherein:
the data acquisition module is used for acquiring logistics data of each logistics network from the block chain platform;
the model construction module is used for constructing a default risk assessment model based on the acquired logistics data of each logistics network;
and the evaluation module is used for evaluating the default risk of each logistics network point through the default evaluation model so as to obtain the default risk score of each logistics network point.
4. The block chain-based logistics information platform default risk management and control system of claim 3, wherein the logistics data comprises historical order data, historical credit characteristics, people credit characteristics and real-time business data, wherein the historical order data comprises formatted order data and unformatted order data;
the step of constructing a default risk assessment model based on the logistics data of each logistics network comprises the following steps:
extracting first default risk data characteristics related to default from the formatted order data by adopting a random forest algorithm;
extracting a second type of default risk data characteristics related to default from the unformatted order data by adopting a deep learning algorithm;
and constructing a default risk assessment model based on the first default risk data characteristic, the second default risk data characteristic, the historical credit characteristic, the character credit characteristic and the real-time business data of each logistics network point.
5. The block chain-based logistics information platform default risk management and control system of claim 4, wherein the extracting of the first type of default risk data features related to the default from the formatted order data by using a random forest algorithm comprises:
acquiring a sample data set comprising a plurality of formatted order data samples;
randomly extracting m training samples from the sample data set in a place where the sample data set is placed back by using a bootstrapping sampling method, and performing n rounds of extraction to obtain n training sets;
respectively training n decision tree models based on n training sets;
respectively calculating the importance value of each feature of the formatted order data by using n decision tree models, and averaging the n importance values of each feature to obtain a unique determined importance value of each feature;
selecting a plurality of characteristics of which the importance value is determined to exceed a predetermined threshold value as the first type of default risk data characteristics.
6. The block chain-based logistics information platform breach risk management and control system of claim 4, wherein the extracting a second type breach risk data feature related to breach from the unformatted order data by using a deep learning algorithm comprises:
converting the unformatted order data into a plurality of word vectors;
constructing a feature matrix based on the word vectors;
and performing feature extraction on the feature matrix by adopting a convolutional neural network to obtain the second-class default risk data features.
7. The block chain-based logistics information platform default risk management and control system of claim 4, wherein: the constructing a default risk assessment model based on the first type of default risk data characteristics, the second type of default risk data characteristics, the historical credit characteristics, the people credit characteristics, and the real-time business data comprises:
a component graph neural network G ═ V, E, where V is a set of nodes and E is a set of edges;
adding each logistics network point into a node set as a node, wherein the first type default risk data characteristic, the second type default risk data characteristic, the historical credit characteristic, the character credit characteristic and the real-time service data of each logistics network point are used as the data characteristics of the corresponding node;
and acquiring a logistics transportation record from the block chain platform to acquire the relation among the logistics network points, adding the acquired relation among the logistics network points into the edge set, and acquiring the default risk assessment model based on the graph neural network.
8. The block chain-based logistics information platform default risk management and control system of claim 7, wherein: the step of evaluating the default risk of each logistics network point through the default evaluation model to obtain the default risk score of each logistics network point comprises:
selecting nodes with default risk labels as seed nodes, and executing a random walk algorithm in the default risk evaluation model by adopting a Node2VEC algorithm to obtain the sampling probability of each Node;
and obtaining the corresponding risk score of each logistics network point based on the sampling probability of each node.
9. The block chain-based logistics information platform default risk management and control system of claim 1, wherein the classifying the default risk management and control of each logistics site based on the default risk score of each logistics site comprises:
for the first type of logistics network points with default risk scores lower than a preset threshold value, adopting a first type of risk control measures;
and taking a second type of risk control measures for the second type of logistics network points with default risk scores equal to or higher than a preset threshold value.
10. A logistics information platform default risk control method based on a block chain is characterized by comprising the following steps:
acquiring logistics data uploaded by each logistics network from the block chain platform, wherein the logistics data comprise historical order data, historical credit characteristics, character credit characteristics and real-time service data;
constructing a default risk assessment model based on the logistics data of each logistics network point acquired from the block chain platform, and assessing the default risk of each logistics network point through the default assessment model to obtain a default risk score of each logistics network point;
and acquiring default risk scores of the logistics outlets generated by the default risk assessment system, and performing classified default risk control on the logistics outlets based on the default risk scores of the logistics outlets.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605907.0A CN113379211B (en) | 2021-05-31 | 2021-05-31 | Logistics information platform default risk management and control system and method based on blockchain |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605907.0A CN113379211B (en) | 2021-05-31 | 2021-05-31 | Logistics information platform default risk management and control system and method based on blockchain |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113379211A true CN113379211A (en) | 2021-09-10 |
| CN113379211B CN113379211B (en) | 2024-07-26 |
Family
ID=77575225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110605907.0A Active CN113379211B (en) | 2021-05-31 | 2021-05-31 | Logistics information platform default risk management and control system and method based on blockchain |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113379211B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117237123A (en) * | 2023-11-15 | 2023-12-15 | 山东政信大数据科技有限责任公司 | Financial credit risk intelligent management and control method and system based on blockchain |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170124496A1 (en) * | 2015-10-28 | 2017-05-04 | Samsung Sds Co., Ltd. | Risk scoring method in logistics system |
| CN110287180A (en) * | 2019-06-25 | 2019-09-27 | 上海诚数信息科技有限公司 | A kind of air control modeling method based on deep learning |
| CN110415111A (en) * | 2019-08-01 | 2019-11-05 | 信雅达系统工程股份有限公司 | Merge the method for logistic regression credit examination & approval with expert features based on user data |
| CN110472040A (en) * | 2019-06-26 | 2019-11-19 | 平安科技（深圳）有限公司 | Extracting method and device, storage medium, the computer equipment of evaluation information |
| CN110599098A (en) * | 2019-09-17 | 2019-12-20 | 上海天地汇供应链科技有限公司 | Logistics industry-based block chain credit investigation management method and system |
| CN111324862A (en) * | 2020-02-10 | 2020-06-23 | 深圳华策辉弘科技有限公司 | Method and system for monitoring behavior in loan |
| CN111860783A (en) * | 2020-07-22 | 2020-10-30 | 中山大学 | Graph node low-dimensional representation learning method and device, terminal equipment and storage medium |
| CN112184012A (en) * | 2020-09-27 | 2021-01-05 | 平安资产管理有限责任公司 | Enterprise risk early warning method, device, equipment and readable storage medium |
| US20210065188A1 (en) * | 2019-08-29 | 2021-03-04 | International Business Machines Corporation | Implementing a marketplace for risk assessed smart contracts issuers and execution providers in a blockchain |
| KR20210060982A (en) * | 2019-11-19 | 2021-05-27 | 인포뱅크 주식회사 | A Cryptographic liquidity borrowing method and a system using block chain with default resistance |
-
2021
- 2021-05-31 CN CN202110605907.0A patent/CN113379211B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170124496A1 (en) * | 2015-10-28 | 2017-05-04 | Samsung Sds Co., Ltd. | Risk scoring method in logistics system |
| CN110287180A (en) * | 2019-06-25 | 2019-09-27 | 上海诚数信息科技有限公司 | A kind of air control modeling method based on deep learning |
| CN110472040A (en) * | 2019-06-26 | 2019-11-19 | 平安科技（深圳）有限公司 | Extracting method and device, storage medium, the computer equipment of evaluation information |
| CN110415111A (en) * | 2019-08-01 | 2019-11-05 | 信雅达系统工程股份有限公司 | Merge the method for logistic regression credit examination & approval with expert features based on user data |
| US20210065188A1 (en) * | 2019-08-29 | 2021-03-04 | International Business Machines Corporation | Implementing a marketplace for risk assessed smart contracts issuers and execution providers in a blockchain |
| CN110599098A (en) * | 2019-09-17 | 2019-12-20 | 上海天地汇供应链科技有限公司 | Logistics industry-based block chain credit investigation management method and system |
| KR20210060982A (en) * | 2019-11-19 | 2021-05-27 | 인포뱅크 주식회사 | A Cryptographic liquidity borrowing method and a system using block chain with default resistance |
| CN111324862A (en) * | 2020-02-10 | 2020-06-23 | 深圳华策辉弘科技有限公司 | Method and system for monitoring behavior in loan |
| CN111860783A (en) * | 2020-07-22 | 2020-10-30 | 中山大学 | Graph node low-dimensional representation learning method and device, terminal equipment and storage medium |
| CN112184012A (en) * | 2020-09-27 | 2021-01-05 | 平安资产管理有限责任公司 | Enterprise risk early warning method, device, equipment and readable storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117237123A (en) * | 2023-11-15 | 2023-12-15 | 山东政信大数据科技有限责任公司 | Financial credit risk intelligent management and control method and system based on blockchain |
| CN117237123B (en) * | 2023-11-15 | 2024-03-01 | 山东政信大数据科技有限责任公司 | Financial credit risk intelligent management and control method and system based on blockchain |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113379211B (en) | 2024-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110223168B (en) | Label propagation anti-fraud detection method and system based on enterprise relationship map | |
| CN110516910A (en) | Declaration form core based on big data protects model training method and core protects methods of risk assessment | |
| CN114066242A (en) | Enterprise risk early warning method and device | |
| CN106067094A (en) | A kind of dynamic assessment method and system | |
| CN113051291A (en) | Work order information processing method, device, equipment and storage medium | |
| CN113554310A (en) | Enterprise credit dynamic evaluation model based on intelligent contract | |
| CN110619462A (en) | Project quality assessment method based on AI model | |
| CN111612603A (en) | Suspected financial counterfeiting behavior insights and discrimination system of listed company | |
| CN117726166A (en) | Artificial intelligence enterprise customer risk information analysis and evaluation method and system based on large language model | |
| CN114612239A (en) | Stock public opinion monitoring and wind control system based on algorithm, big data and artificial intelligence | |
| Jiang et al. | [Retracted] Research on Intelligent Prediction Method of Financial Crisis of Listed Enterprises Based on Random Forest Algorithm | |
| CN113379211A (en) | Block chain-based logistics information platform default risk management and control system and method | |
| CN117114812A (en) | Financial product recommendation method and device for enterprises | |
| CN117236698A (en) | Industrial chain risk assessment model, method and medium based on graph neural network | |
| CN113379212B (en) | Logistics information platform default risk assessment method, device, equipment and medium based on blockchain | |
| Jeyaraman et al. | Practical Machine Learning with R: Define, build, and evaluate machine learning models for real-world applications | |
| CN116304929A (en) | Financial manipulation recognition method and device based on A-stock market | |
| KR102499182B1 (en) | Loan regular auditing system using artificia intellicence | |
| Khalifa et al. | Fake reviews detection based on both the review and the reviewer features under belief function theory | |
| Herreros-Martínez et al. | Applied Machine Learning to Anomaly Detection in Enterprise Purchase Processes | |
| CN118710283A (en) | Customer service capability portrait generating method, device and medium | |
| CN117745409A (en) | Block chain-based industrial chain financial wind control model construction method | |
| HEDENBERG | Developing components for a data-driven trading system | |
| CN117973846A (en) | Enterprise risk prediction method and system based on industrial chain | |
| Tang et al. | Exploring assessment method of technological advancement based on literature cross-citation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |